The (Greek) Key to Structures of Neural Adhesion Molecules Daniel E Vaughn, Pamela J Bjorkman  Neuron  Volume 16, Issue 2, Pages 261-273 (February 1996) DOI: 10.1016/S0896-6273(00)80045-8

Figure 1 Domain Organization of Neural Ig Superfamily CAMs Ig-like and Fn-III domains are indicated for the extracellular regions of vertebrate CAMs; structural homologs in insects are listed in parentheses. The domains are classified based upon a comparison of their sequences to the structure-based sequence alignments in Figure 2. No classification is listed for those domains that do not show a clear agreement to one of the consensus sequences. Neuron 1996 16, 261-273DOI: (10.1016/S0896-6273(00)80045-8)

Figure 1 Domain Organization of Neural Ig Superfamily CAMs Ig-like and Fn-III domains are indicated for the extracellular regions of vertebrate CAMs; structural homologs in insects are listed in parentheses. The domains are classified based upon a comparison of their sequences to the structure-based sequence alignments in Figure 2. No classification is listed for those domains that do not show a clear agreement to one of the consensus sequences. Neuron 1996 16, 261-273DOI: (10.1016/S0896-6273(00)80045-8)

Figure 2 Structure-Based Sequence Alignments Structural alignment of residues is based upon the pairwise superposition of V-like and C2 domains (Table 1Table 2 ). Gaps in one sequence compared with the others are indicated by dots. β strands are colored red (ABE-containing sheet) or blue (GFC-containing sheet). Structurally conserved loops characteristic of V-like domains (connecting strands A′ to B and E to F) are colored yellow. Sequences between strands C and D are not listed in the Ig-like domain sequences (indicated by double slashes and colored green) because the high degree of variability between structures makes alignment impossible. Consensus primary sequence patterns are identified at the bottom of the sequences: an asterisk indicates a hydrophobic amino acid; a plus sign represents a basic amino acid; a number sign indicates a glycine, alanine, or aspartate; and an “x” indicates any amino acid. Neuron 1996 16, 261-273DOI: (10.1016/S0896-6273(00)80045-8)

Figure 4 Tandem Domain Interfaces in CAMs Ribbon diagrams are shown for tandem domain structures. The ABED sheet is red and the A′GFCC′C′′ sheet is blue. In the two diagrams of CD4, the strand that continues from the first domain into the second domain is highlighted in green. For the CD2 and VCAM-1 diagrams, the short interdomain connecting sequence is highlighted in green. A metal ion between the neuroglian Fn-III domains (NgFn1,2) is shown in green. The integrin binding loop in VCAM-1 and residues within the metal binding site in NgFn1,2 are highlighted in yellow. These figures were prepared with Molscript (Kraulis 1991) and rendered with Raster 3D (Merritt and Murphy 1994) from coordinates obtained from the protein database (3cd4 for CD4 D1,2; 1cid for CD4 D3,4; 1hgf for CD2, and 1cfb for NgFn1,2) or provided by E. Y. Jones for VCAM-1. Neuron 1996 16, 261-273DOI: (10.1016/S0896-6273(00)80045-8)

Figure 3 Structures of Building Block Domains The border separating diagrams of antibody domains (top) and from neural CAM domains (bottom) shows a typical Greek key pattern as seen in vases and other early Greek art. Topology (above the fold name) and ribbon (below the fold name) diagrams are presented for each building block structure. The ABE-containing sheets are red and the GFC-containing sheets are blue. In the topology diagrams, strands are identified by letters. Amino acids that have equivalent positions in all structures of the domain type are indicated by a closed circle, by the one-letter code if the identity of the amino acid is conserved, by the symbol φ for hydrophobic residues, or by the symbol ψ for hydrophilic residues. β-sheet hydrogen bonding is indicated by dashed lines. Regions of irregular secondary structure are indicated by open rectangles. Yellow highlights the antigen binding loops in the Ig variable domain, the structurally conserved loops in the V-like domains (A′ to B and E to F loops), the integrin-binding RGD loop present in some Fn-III domains, and the A strand of N-cadherin domain 1 that mediates formation of the strand dimer. The part of the V-like domain that shows the most variability between structures (connection between strands C and D) is highlighted in green. Ribbon diagrams were prepared using Molscript (Kraulis 1991) and rendered with Raster 3D (Merritt and Murphy 1994) from coordinates available from the PHB (7fab for Ig constant and variable domains, 1ten for Fn-III) or provided by the authors (VCAM-1 coordinates from E. Y. Jones for V-like and C2 domains, and N-cadherin domain 1 coordinates from L. Shapiro for Cad). Neuron 1996 16, 261-273DOI: (10.1016/S0896-6273(00)80045-8)

Figure 5 Hypothetical Model for the Alignment of Six Ig-like and Five Fn-III Domains in Series, as in Molecules of the L1 Family Space-filling model in which atoms in Ig-like domains are blue (side-chain atoms) and green (main-chain atoms). Main-chain atoms in Neuron 1996 16, 261-273DOI: (10.1016/S0896-6273(00)80045-8)

Figure 6 Models of Homophilic Adhesive Interactions for Ig V-like and Cadherin Domains (Left) Ribbon diagram of the head-to-head interaction observed in crystals of CD2 domains 1 and 2 (Jones et al. 1992) serves as a model for homophilic and heterophilic interaction between Ig superfamily members. (Right) The zipperlike structure observed in crystals of N-cadherin domain 1 (Shapiro et al. 1995a) as a model for cadherin-mediated cell–cell adhesion. Strand A, which interdigitates into the partner domain of the strand dimer, is highlighted in yellow. Neuron 1996 16, 261-273DOI: (10.1016/S0896-6273(00)80045-8)